Sparse Coding
How biological networks represent complex concepts through highly specific, distributed patterns rather than dense, continuous blobs.
- Sparse distributed representations
- Sparse activity
- Pattern overlap
- Interpretability potential
Predictive Processing
The theory that the brain is fundamentally a prediction engine, constantly minimizing the delta between expectation and sensory reality.
- Prediction error
- Top-down and bottom-up signals
- Local learning inspiration
Working Memory and Recurrence
How biological intelligence maintains persistent state over time, allowing for reasoning across temporal gaps.
- State over time
- Gated recurrence
- Memory update and forgetting
Relational Binding
The ability to dynamically map entities to roles, allowing for compositionality and structural understanding.
- Binding entities, roles, and relations
- Structured internal representations
What We Are Not Claiming
- We are not claiming that CMP replicates cortical circuits.
- We are not claiming biological accuracy.
- We are using neuroscience to guide engineering hypotheses.